SUPPORTING INFORMATION

Transcription

1 SUPPORTING INFORMATION Copper Silver Thin Films with Metastable Miscibility for Oxygen Reduction Electrocatalysis in Alkaline Electrolytes Drew Higgins, 1,2 Melissa Wette, 3 Brenna M. Gibbons, 2,3 Samira Siahrostami, 1,2 Christopher Hahn, 1,2 Maria Escudero-Escribano, 1,4 Max García-Melchor, 1,2,5 Zachary Ulissi, 1,2 Ryan C. Davis, 6 Apurva Mehta, 6 Bruce M. Clemens*, 3 Jens K. Nørskov*, 1,2 Thomas F. Jaramillo* 1,2 1 Department of Chemical Engineering, Stanford University, 443 Via Ortega Way, Stanford, California, 94305, USA 2 SUNCAT Center for Interface Science and Catalysis, SLAC National Accelerator Laboratory, 2575 Sand Hill Road, Menlo Park, California, 94025, USA 3 Department of Materials Science and Engineering, Stanford University, 496 Lomita Mall, Stanford, California, 94305, USA 4 Department of Chemistry, University of Copenhagen, Universitetsparken 5, 2100 Copenhagen, Denmark 5 School of Chemistry, Trinity College Dublin, College Green, Dublin 2, Ireland 6 Stanford Synchrotron Radiation Lightsource, SLAC National Accelerator Laboratory, 2575 Sand Hill Road, Menlo Park, California, 94025, USA Corresponding author s: jaramillo@stanford.edu; norskov@stanford.edu; bmc@stanford.edu S-1

2 Specific (Surface Area Based) Activity Comparison Table S1 -Turn over frequencies on a surface area basis for state of the art Pt/HSC, Cu 70 Ag 30 thin films, and a sample of best-performing Fe-N-C catalysts (that report BET surface areas and mass loadings) Material i 0.8 V vs RHE Units Reference a Pt/HSC 6.5 ma/cm 2 Pt [1] Ag(110) 0.78 Ag(111) 0.26 Ag(100) 0.20 ma/cm 2 Ag [2] b Cu 70 Ag 30 Thin Film 1.62 ma/cm 2 electrode This work c Fe-N-C 0.08 ma/cm 2 Fe-N-C [3] 0.09 [4] a Pt/HSC = State of the art high surface area carbon supported platinum (46 wt% Pt, TKK, TEC10E50E); data is extracted from Figure 13 of this reference for a highly optimized electrode preparation, and activity reported in acidic (0.1 M HClO 4 ) electrolyte. This is used as a representative case as it is wellknown that, in the absence of electrolyte impurities, Pt catalysts provide very similar activity between acid and alkaline electrolytes. b Attempts were made to quantify the surface compositions of Ag and Cu under electrochemical reaction conditions using Pb underpotential deposition (Pb-upd).[5] These attempts were unsuccessful due to Ag and Cu both exhibiting Pb-upd features in the same potential range, which could not be deconvoluted. Activity values are therefore reported on an electrode (geometric) area basis. These attempts also indicate that there is opportunity for scientific advancement through the development of new techniques to quantify surface areas of different elements in alloy catalysts under relevant electrochemical conditions. c Fe-N-C values are estimated based on available electrode catalyst loadings, BET surface areas and Tafel plots. In the case of [4], the low overpotential linear portion of the Tafel plot was extrapolated to 0.8 V vs RHE, which gives a best case turn over frequency. S-2

3 Vegard s Law Analysis Vegard s law states that the lattice parameter of a single phase, two-component system is a linear combination of the lattice parameters of each of the components.[6] For the copper-silver bimetallic system, this can be expressed as follows: a Cux Ag 1 x = 0.01(xa Cu + (1 x)a Ag ) where x is the nominal atomic percent of copper, and a is the lattice parameter of some phase i. We extract the lattice parameters from the symmetric scan XRD patterns in Figure 1 via Bragg s law and calculate the percent copper and silver in each phase present in Table S1 below. The two-phase nature of Cu 50 Ag 50 and Cu 70 Ag 30 can be plainly seen by the presence of two sets of (111) peaks in these diffraction data. Cu 20 Ag 80 and Cu 90 Ag 10 both appear to be single-phase by the presence of only (111) reflection in each scan; however, closer examination shows this is not the case. Vegard s law analysis indicates that the Ag-rich phase in Cu 20 Ag 80 is only 13% Cu, which implies that there must be another Curich phase that contains the rest of the sample s 20% copper content. This signal from this minority Curich phase may be too small to be detected above the noise. Though copper and silver have almost no solubility in one another at room temperature, PVD thin film synthesis allows us to freeze in metastable solid solution phases, here with up to 21.8% copper and 28.6% silver content, respectively. Of the four bimetallic compositions prepared, only the most copperrich Cu 90 Ag 10 is truly single-phase. Table S2 Vegard s law analysis of XRD patterns showing interphase miscibility and strain. Sample Peak 2θ (degrees) a (Å) % Cu % Ag % Strain Cu 90 Ag 10 Cu Cu Cu 70 Ag 30 Ag Cu Cu 50 Ag 50 Ag Cu Cu 20 Ag 80 Ag S-3

4 Additional Characterization Figure S1 - SEM of Cu, CuAg and Ag thin films (left) as-prepared and (right) after ORR measurements. S-4

5 Figure S2 - Cyclic voltammetry at 50 mv/s in Ar-sparged 0.1 M KOH for (a) Cu 90 Ag 10, (b) Cu 50 Ag 50, (c) Cu 20 Ag 80 and (d) Ag thin film electrode. S-5

6 Figure S3 - AFM imaging of Cu 70 Ag 30 electrode after ORR measurement. S-6

7 Figure S4 - XPS survey spectra and high-resolution scans for (black, bottom) as prepared and (red, top) electrochemically ORR tested Cu 70 Ag 30 thin film electrodes. S-7

8 Figure S5 - XPS sputter depth profile of Cu 70 Ag 30 thin film electrode. Approximate sputter rate of 4 nm/s. 1kV, 0.5uA, 2x2 um. S-8

9 Limiting Potential / V Strain Effects Compressive 0-5% Tensile 0-5% Strain % Figure S6 - Strain effect on the calculated limiting potentials of Ag-rich phase. S-9

10 Limiting Potential / V Cu/Ag Miscibility Table S3 Binding energies of oxygen containing adsorbates on Cu a in Cu a -Ag model surfaces Configuration G OH (ev) G OOH (ev) G O (ev) Cu a -Ag Cu a -Ag(-2%) Cu a -Ag(+2%) Compressive 0-5% Tensile 0-5% Strain % Figure S7 - Strain effect on the calculated limiting potentials of Cu atom trapped in Ag rich phase. S-10

11 Projected Density of States Analysis To understand the reactivity of different Cu a -Ag surfaces we use the d-band model, which has been used extensively for metal surfaces and shows a correlation with adsorbate binding energies.[7-9] Figure S9a displays the d-projected density of states (PDOS) for a Cu atom in pure Cu(111), and for a Cu atom in Cu a -Ag with 0%, +2% (tensile) and -2% (compressive) strain versus pure Ag. As seen from this figure, the electronic structure of the single Cu atom in Cu a -Ag is substantially different from a Cu atom in Cu(111). For Cu in -2% compressively strained Ag, namely Cu a -Ag(-2%), a lower d-band edge relative to the Fermi level is observed relative to all of the Cu a -Ag configurations investigated. This leads to a weaker oxygen binding energy for Cu and resultant increase in ORR activity versus unstrained or tensile strained Cu a -Ag, in addition to pure Cu. On this basis, theoretical calculations suggest that the experimentally observed high activity for Cu 70 Ag 30 could arise from the number of Cu atoms trapped in the compressively strained Ag rich phase under ORR conditions. On the other hand, Figure S9b shows a similar electronic structure for Ag atoms in Cu a -Ag versus Ag(111), indicating that the ORR activity of these Ag atoms should not be adversely affected by the neighbouring Cu. We have also investigated the Ag a -Cu surface, with the bridge site of an Ag atom (Figure S9c) found to have a limiting potential of 0.49 V vs RHE. This indicates that the Ag atoms trapped within the Cu-rich phase can still contribute to ORR activity, albeit likely are not underlying the significant ORR enhancements observed in comparison to pure Ag. Figure S8 - d-projected density of states for the model systems of (a) Cu atom in Cu(111), Cu a -Ag with - 2%, 0% and +2% strain (b) Ag atom in Ag(111), Cu a -Ag with -2%, 0% and +2% strain. (c) Ag a -Cu surface with bridge site indicated by adsorbate. S-11

12 Cu-oxide or Ag Overlayers on Cu-oxide The presence of Cu-oxide species under ORR conditions (Figure 2a) required investigation of CuO and Cu 2 O on activity. For pure oxides, we investigated the (100) facet of CuO due to its highest activity among low index facets[10], and for Cu 2 O we investigated the three low index facets, namely (111), (110) and (100). All examined Cu 2 O surfaces are completely inactive towards the ORR due to the strong interactions with oxygenated species that poison the surface. The CuO(100) facet has a calculated limiting potential of 0.47 V vs RHE (Figure 5b, Figure S7), albeit this phase is generally not formed until above ca V vs RHE and provides insulating behaviour that could explain the experimentally observed lack of ORR activity on pure Cu in an oxidized state (Figure 3b). Because XPS and GI-XRD characterization of post ORR CuAg electrodes indicated a surface predominantly comprised of an Ag-rich phase, we have also considered the possibility of an Ag overlayer on top of different Cu-oxide substrates (Figure 5a). Ag overlayers on CuO and Cu 2 O undergo tensile strain due to the larger lattice constants of the oxide structures, with the Cu 2 O(100) and (110) substrates excluded due to extensive strain (+9% and +17%, respectively) that leads to major surface reconstruction. Conversely, reasonably stable Ag overlayers can form on CuO(100) and Cu 2 O(111), undergoing tensile strains of 0.4% and 4%, respectively. Figure 5b and Figure S7 shows that an Ag overlayer on CuO(100) or Cu 2 O(111) exhibit calculated limiting potentials of 0.60 and 0.40 V vs RHE, respectively. DFT analysis therefore suggests that the ORR activity enhancement does not arise directly from an active Cu-oxide species, but potentially due to an Ag overlayer present on the surface. Figure S9 - Calculated limiting potentials on the CuO(100) and Cu2O(111), (110), (100) and their corresponding Ag overlayers. S-12

13 References (1) Shinozaki, K., Zack, J.W., Pylypenko, S., Pivovar, B.S., Kocha, S.S., Oxygen Reduction Reaction Measurements on Platinum Electrocatalysts Utilizing Rotating Disk Electrode Technique: II. Influence of Ink Formulation, Catalyst Layer Uniformity and Thickness. Journal of The Electrochemical Society, 2015, 162, F1384. (2) Blizanac, B.B., Ross, P.N., Marković, N.M., Oxygen Reduction on Silver Low-Index Single-Crystal Surfaces in Alkaline Solution: Rotating Ring DiskAg(hkl) Studies. The Journal of Physical Chemistry B, 2006, 110, (3) Qiu, K., Chai, G., Jiang, C., Ling, M., Tang, J., Guo, Z., Highly Efficient Oxygen Reduction Catalysts by Rational Synthesis of Nanoconfined Maghemite in a Nitrogen-Doped Graphene Framework. ACS Catalysis, 2016, 6, (4) Xia, B.Y., Yan, Y., Li, N., Wu, H.B., Lou, X.W., Wang, X., A metal organic framework-derived bifunctional oxygen electrocatalyst. Nature Energy, 2016, 1, (5) Herrero, E., Buller, L.J., Abruña, H.D., Underpotential Deposition at Single Crystal Surfaces of Au, Pt, Ag and Other Materials. Chemical Reviews, 2001, 101, (6) Denton, A.R., Ashcroft, N.W., Vegard's law. Physical Review A, 1991, 43, (7) Hammer, B., Norskov, J.K., Why gold is the noblest of all the metals. Nature, 1995, 376, 238. (8) Mavrikakis, M., Hammer, B., Nørskov, J.K., Effect of Strain on the Reactivity of Metal Surfaces. Physical Review Letters, 1998, 81, (9) Kitchin, J.R., Nørskov, J.K., Barteau, M.A., Chen, J.G., Role of Strain and Ligand Effects in the Modification of the Electronic and Chemical Properties of Bimetallic Surfaces. Physical Review Letters, 2004, 93, (10) Su, D., Xie, X., Dou, S., Wang, G., CuO single crystal with exposed {001} facets - A highly efficient material for gas sensing and Li-ion battery applications. 2014, 4, S-13